feat(fr): bis/ter housenumber suffixes

pelias · Sep 9, 2024 · e98ff9c · e98ff9c
1 parent 14c5be7
commit e98ff9c
Show file tree

Hide file tree

Showing 5 changed files with 42 additions and 4 deletions.
diff --git a/classifier/scheme/subdivision.js b/classifier/scheme/subdivision.js
@@ -0,0 +1,19 @@
+const HouseNumberClassification = require('../../classification/HouseNumberClassification')
+
+module.exports = [
+  {
+    // 10 bis / 10 ter
+    confidence: 0.99,
+    Class: HouseNumberClassification,
+    scheme: [
+      {
+        is: ['HouseNumberClassification'],
+        not: ['IntersectionClassification']
+      },
+      {
+        is: ['StopWordClassification'],
+        not: ['IntersectionClassification', 'PunctuationClassification']
+      }
+    ]
+  }
+]
diff --git a/parser/AddressParser.js b/parser/AddressParser.js
@@ -78,6 +78,7 @@ class AddressParser extends Parser {
         new CompositeClassifier(require('../classifier/scheme/street')),
         new CompositeClassifier(require('../classifier/scheme/venue')),
         new CompositeClassifier(require('../classifier/scheme/intersection')),
+        new CompositeClassifier(require('../classifier/scheme/subdivision')),
 
         // additional classifiers which act on unclassified tokens
         new CentralEuropeanStreetNameClassifier()

diff --git a/resources/pelias/dictionaries/libpostal/fr/stopwords.txt b/resources/pelias/dictionaries/libpostal/fr/stopwords.txt
@@ -0,0 +1,2 @@
+bis
+ter
diff --git a/test/address.fra.test.js b/test/address.fra.test.js
@@ -117,6 +117,22 @@ const testcase = (test, common) => {
   assert(`Esplanade de la Liberté`, [{ street: 'Esplanade de la Liberté' }])
   assert(`Esplanade du Géneral de Gaulle`, [{ street: 'Esplanade du Géneral de Gaulle' }])
   assert(`Esplanade Méditerranée`, [{ street: 'Esplanade Méditerranée' }])
+
+  // bis/ter housenumber prefixes
+  assert(`1 bis Av. Amélie, 92320 Châtillon, France`, [
+    { housenumber: '1 bis' },
+    { street: 'Av. Amélie' },
+    { postcode: '92320' },
+    { locality: 'Châtillon' },
+    { country: 'France' }
+  ])
+  assert(`1 ter Av. Amélie, 92320 Châtillon, France`, [
+    { housenumber: '1 ter' },
+    { street: 'Av. Amélie' },
+    { postcode: '92320' },
+    { locality: 'Châtillon' },
+    { country: 'France' }
+  ])
 }
 
 module.exports.all = (tape, common) => {

diff --git a/test/address.usa.test.js b/test/address.usa.test.js
@@ -217,23 +217,23 @@ const testcase = (test, common) => {
   ]], false)
 
   // NYC Boroughs
-  assert('866 E 178th St, Bronx, NY 10460, USA', [[
+  assert('866 E 178th St, Bronx, NY 10460, USA', [
     { housenumber: '866' },
     { street: 'E 178th St' },
     { locality: 'Bronx' },
     { region: 'NY' },
     { postcode: '10460' },
     { country: 'USA' }
-  ]], false)
+  ])
 
-  assert('866 E 178th St, Staten Island, NY 10460, USA', [[
+  assert('866 E 178th St, Staten Island, NY 10460, USA', [
     { housenumber: '866' },
     { street: 'E 178th St' },
     { locality: 'Staten Island' },
     { region: 'NY' },
     { postcode: '10460' },
     { country: 'USA' }
-  ]], false)
+  ])
 
   // 'Massachusetts' and 'MA' should be interchangeable and both
   // forms should allow 'Boston' to be parsed as a locality.